Introduction

In recent decades, rising CO2 emissions have posed significant threats to global sustainability and climate stability (Calvin et al. 2023). CO2 emissions, accounting for approximately 76% of total greenhouse gas emissions, increased by 28.67% between 2003 and 2018 (Friedlingstein et al. 2019; Janssens-Maenhout et al. 2019), prompting widespread adoption of emission-reduction targets. The COP28 summit emphasized the critical need for $3.8 trillion in annual investments to address climate challenges, particularly in emerging economies (Shah 2023). The E-7 economies face a dual challenge of sustaining rapid economic growth while mitigating environmental degradation from their historic fossil fuel-dependent industrial expansion (Chen et al. 2024; Gao et al. 2024). These economies have increasingly prioritized sustainable development, implementing energy efficiency measures and transitioning toward renewable energy. However, achieving carbon neutrality requires not only renewable energy but also strategic financial inclusion policies that channel resources toward sustainable investments rather than carbon-intensive activities (Gao et al. 2024). This nexus between financial inclusion and renewable energy adoption represents a critical pathway for emerging economies to reconcile development objectives with climate commitments.

The renewable energy strategy in the emerging seven economies, particularly in China, demonstrates that energy efficiency and renewable energy policies have significant implications for human livelihoods, economic performance, and energy security (Lo 2014). Renewable energy has become more crucial due to climate change and worries about long-term environmental health (Huang et al. 2024; Raihan and Tuspekova 2022). The introduction of digital financial inclusion and changing old ways by using digital technology could influence energy consumption patterns, industrial energy ecosystems, and carbon emissions (Fu et al. 2024; Iqbal et al. 2022; Lee et al. 2022). Renzhi and Baek (2020) showed a negative link between CO2 and financial inclusion. They discovered that the relationship between financial inclusion and carbon emissions is an inverted U-shaped. Therefore, reaching a high degree of financial inclusion will compel investors to embrace tougher policies that enhance environmental quality, hence reducing carbon emissions. Figure 1 gives the financial inclusion trends for E-7 economies from 2000 to 2023, showing an upward trajectory across all countries, with China and India showing particularly robust growth, while Russia experienced a decline after peaking in 2014. This expansion of financial services access across E-7 nations potentially enhances their capacity to channel investments toward renewable energy projects.

Fig. 1
figure 1

Financial inclusion. Source: Author’s own calculation, index created using PCA analysis (see “Methodology” section).

Full size image

This study investigates the renewable energy-emissions nexus in E-7 economies, motivated by several critical factors. First, while theoretical and empirical research generally confirms that renewable energy reduces CO2 emissions (Dogan and Seker, 2016), the magnitude and effectiveness of this relationship vary significantly across different economic contexts and development stages, with limited understanding of the moderating factors that explain this heterogeneity. Second, existing literature has largely overlooked the moderating role of financial inclusion in the renewable energy-emissions relationship, particularly in emerging economies where rapid financial sector expansion coincides with energy transition decisions. Third, the E-7 economies present a unique research context as they collectively account for over 50% of global CO2 emissions (Friedlingstein et al. 2025) while experiencing diverse stages of financial development and renewable energy adoption, making them ideal for examining heterogeneous policy effects through quantile regression analysis. Fourth, these economies face the dual challenge of sustaining rapid economic growth while meeting climate commitments under the Paris Agreement, necessitating evidence-based insights into how financial development patterns may either support or undermine renewable energy effectiveness. Lastly, the increasing urgency of climate action, as emphasized at COP28 with calls for higher annual climate investments (Shah, 2023), underscores the need to understand how financial inclusion mechanisms can be aligned with renewable energy deployment to achieve meaningful emission reductions in the world’s most climate-critical economies. Figure 2 presents a bar chart comparing renewable energy (RE) and financial inclusion (FI) metrics across E-7 economies, where bar width represents RE values and height represents FI values. The chart reveals substantial heterogeneity, with Brazil and China exhibiting the highest renewable energy capacities, Turkey demonstrating financial inclusion with moderate RE levels, and Russia showing relatively low values for both metrics.

Fig. 2
figure 2

RE & FI. Source: Author’s own calculation, index created using PCA analysis (see “Methodology” section), and RE data taken from World Bank.

Full size image

As evident from the above discussion, RE plays an important role in the sustainable development of emerging countries. Therefore, this study explores the impact of financial inclusion (FI) on the relationship between renewable energy (RE) and trade-adjusted carbon emissions. The present study also assesses the role of FI in carbon emissions, as it has received limited attention in the E-7 economies. The present study adds to the literature and differs from earlier studies by analyzing the role of FI and RE and taking the newly developed consumption-based carbon emissions (CCO2) metric. The newly developed metric incorporates fossil fuels emissions and carbon emissions resulting from international trade (both imports and exports), addressing a previously overlooked aspect in prior studies that solely focused on territorial-based CO2 emissions and disregarded emissions associated with imports and exports. This study is unique from prior research that mostly concentrated on industrialized nations. Instead, our focus is to investigate these seven emerging economies that have shown strong progress over the last two decades and are anticipated to surpass the G-7 economies (PWC, 2017). Due to their significant consumption of 40% of global energy, it is essential to analyze the influence of renewable energy and financial inclusion in terms of environmental quality. This study offers novel perspectives that might stimulate initiatives aimed at enhancing financial inclusivity to promote renewable energy (RE) adoption and mitigate carbon emissions, particularly in rapidly growing E-7 economies. This study will provide valuable insights for formulating sustainable development policies in economies that are experiencing rapid expansion and environmental issues.

This research is ordered as follows: the “Literature review” section gives the literature survey and hypothesis development, the “Methodology” section of the study explains the methodological strategy and theoretical framework, the “Results” section gives the results, and the “Conclusion and policy recommendations” section gives the conclusion and recommendations.

Literature review

Many research studies have been conducted on the linkage between renewable energy, financial inclusion, and CO2 emissions. These studies have produced contradictory results regarding the role of financial inclusion in sustainable development. The following provides a survey of the existing literature.

Financial inclusion (FI) and emissions

According to Mehmood (2020), “financial inclusion means the access of individuals to their financial requirements like transactions, payments, credits, and remittances”. Financial inclusion (FI) is indicative of a nation’s economic stability; however, high levels of economic activity result in increased energy consumption, which has an adverse effect on the environment. FI has a detrimental impact on carbon emissions in the provinces of China (Qayyum 2022). Le et al. (2020) showed that there is an inverted U-shaped linkage between FI and emissions, indicating the existence of EKC in the relationship. Shahbaz et al. (2022) highlighted that the correlation between FI and CO2 emissions is not uniform and differs across different geographical locations. Additionally, environmental degradation is higher at the early stages of economic development. According to Ozili (2020), FI is directly linked to economic development and inequality. Conversely, Renzhi and Baek (2020) showed a detrimental correlation between financial inclusion and CO2. They discovered that the association between FI and environmental pollution is an inverted U-shape. Moreover, Usman et al. (2020) analyzed financial inclusion in the fifteen largest carbon-emitting countries and showed that financial development had a detrimental effect on environmental degradation. This implies that carbon emissions are diminished by financial development. Using data from Malaysia, Shahbaz et al. (2013) discovered that there is a nonlinear association between financial development and CO2 emissions. Additionally, studies suggest that there is a nonlinear relation between carbon emissions and financial inclusion. Dong et al. (2022) observed an inverted U-shaped nexus between FI and greenhouse gas (GHG) emissions. Based on the above literature survey, we conclude that FI has been found to have a substantial influence on environmental degradation. Based on the above discussion, a few studies show a positive linkage between FI and carbon emissions, while others highlight contrasting findings showing the negative influence on emissions. Therefore, we hypothesize:

H1: There is a linkage between FI and trade-adjusted CO2 emission.

Renewable energy (RE) and environmental degradation

Dogan and Seker (2016) have taken the EU panel data to study the effect of renewable energy (RE) and non-renewable energy on CO2 emissions. The authors worked on the data from 1980 to 2012 and found the validity of the EKC hypothesis. The study revealed that the relationship between RE and carbon emissions is bidirectional, while unidirectional causes \(C{O}_{2}\) emissions from real income. Using data for MENA economies from 1980 to 2015, Charfeddine and Kahia (2019) showed that renewable energy has a slight influence on carbon emissions, indicating the sectors are still weak in these economies. Zoundi (2017) also investigated the linkage between RE and carbon emissions in 25 African economies. The findings indicate that RE has an adverse effect on emissions in the short and long term. Zafar et al. (2019) showed that in G-7 and N-11 countries, the increase in the consumption of renewable energy significantly reduces emissions. According to a study of energy consumption and CO2 emissions in 10 Asian economies from 1980 to 2010, RE has an adverse effect on emissions, while fossil fuel energy increases them (Salim et al., 2017). Hossain et al. (2025) examined the impact of renewable energy on both production and consumption-based CO2 emissions in OECD countries. Their findings show that renewable energy significantly improves environmental quality by reducing emissions, with the effect stronger at higher quantiles of the emissions. Wang et al. (2023a) examined Germany’s environmental quality and found that renewable energy significantly improves environmental outcomes, as measured by both CO2 emissions reduction and the load capacity factor, which comprehensively captures biocapacity and ecological footprint dynamics. Iqbal et al. (2025) demonstrated that renewable energy consumption significantly reduces CO2 emissions across 36 countries, with institutional quality further enhancing this emission-reducing effect, highlighting the importance of combining renewable energy investments with strong institutional frameworks.

As discussed in the literature survey above, strong evidence supports the hypothesis that renewable energy diminishes carbon emissions. The studies indicate that increasing RE is linked with a decline in carbon emissions. Therefore, we hypothesize that:

H2: RE reduces trade-adjusted CO2 emissions and improves environmental quality.

Financial inclusion, renewable energy, and emissions

Few studies have investigated the association between financial inclusion (FI) and renewable energy (RE), and they have determined that renewable energy is substantially enhanced by enhanced financial inclusion and reduces energy poverty (Dogan et al. 2021). Feng et al. (2022) examined the effect of FI on RE in China. The study demonstrated that RE in China was amplified by the rise in ATMs and total insurance coverage. They conclude that FI contributes to the enhancement of REC. In contrast to the above, a rise in financial inclusion enhances buying power, empowering people to buy energy-consuming products. The utilization of these products enhances the demand for energy and, therefore, carbon emissions. Frankel and Romer (1999) argue that financial inclusivity can stimulate economic progress, which in turn leads to an increase in energy demand and CO2 emissions. Frankel and Romer (1999) emphasized that the deployment of solar microgrids is significantly impeded by financial exclusion and restrictions, including a lack of access to capital, limited financial support from governments, and a lack of banking services. The logical reasoning for this concept is that financial services' affordability and accessibility can contribute to the adoption of renewable energy and green technologies, thereby enhancing environmental quality by lowering the energy-intensive products and sources. Based on theoretical literature, financial inclusion could enhance renewable energy’s emission-reducing effect by providing capital access and risk management tools for green investments (Shahbaz et al. 2013; Tamazian et al. 2009). Alternatively, financial inclusion in emerging economies may initially channel resources toward carbon-intensive activities that offer higher short-term returns, potentially undermining renewable energy effectiveness without proper green finance frameworks (Le et al. 2020; Zaidi et al. 2021). Moreover, Ben-Cheikh and Rault (2024) found that financial inclusion increases CO2 emissions in lower-income countries through expanded economic activity, with environmental benefits only materializing at later developmental stages when countries have more robust environmental policies. Given these competing perspectives, the moderation effect of financial inclusion on the renewable energy-emissions nexus remains an empirical question. Therefore, we hypothesize:

H3: FI moderates the linkage between RE and trade-adjusted CO2 emission.

Theoretical framework

The proposed theoretical framework, illustrated in Fig. 3, draws on financial development theory (King and Levine 1993), energy transition (Geels 2002), and the Environmental Kuznets Curve (EKC) (Grossman and Krueger 1991) to examine the complex relationships between financial inclusion, renewable energy, and trade-adjusted CO2 emissions in emerging economies. While financial development theory traditionally posits that well-functioning financial markets efficiently allocate resources, our framework shows a different perspective that enhanced financial inclusion, while facilitating greater access to credit and investment opportunities, may paradoxically increase carbon emissions by enabling expanded consumption and production activities that remain predominantly carbon-intensive in emerging economies. This finding challenges the conventional assumption that financial development automatically supports environmental sustainability (Tamazian et al. 2009).

Fig. 3
figure 3

Theoretical framework.

Full size image

The framework’s central contribution lies in identifying the moderating role of financial inclusion on renewable energy’s emission-reducing potential. Drawing on energy transition theory, we demonstrate that while renewable energy adoption directly reduces emissions through fossil fuel substitution, this beneficial effect is significantly attenuated when financial inclusion expands. The positive interaction effect (β₃) suggests that in the absence of targeted green finance mechanisms, increased financial access channels resources toward traditional, carbon-intensive economic activities rather than supporting the renewable energy transition (Le et al. 2020).

The Environmental Kuznets Curve provides additional theoretical grounding for understanding the heterogeneous effects across the emission distribution. Our framework reveals that the relationships between financial inclusion, renewable energy, and emissions vary significantly across quantiles, with renewable energy showing the strongest emission-reducing effects in lower-emitting countries while becoming less effective in higher-emitting economies. This heterogeneity suggests that the EKC’s prediction of an eventual decoupling of growth from emissions may be contingent on the alignment of financial sector development with climate objectives (Ozturk and Acaravci 2013).

In addition to adding to the literature on sustainable development, this framework offers a new assessment compared to prior studies that often assume financial inclusion inherently supports environmental goals. By synthesizing multiple theoretical perspectives with empirical evidence of the interaction effect, it provides crucial insights into how financial systems may inadvertently undermine clean energy transitions in the absence of appropriate policy frameworks. This integrated approach highlights the urgent need for green finance regulations, differentiated policies based on countries’ emission levels, and mechanisms to ensure that expanding financial access supports rather than hinders renewable energy adoption (Wang et al. 2024; Wei et al. 2025). The framework thus contributes novel insights and helps policymakers to craft appropriate policies to achieve sustainable development goals.

Methodology

This study examines the influence of financial inclusion on the relationship between renewable energy consumption and CO2 emissions using economic growth, energy intensity, and natural resources as control variables, taking the E-7 economies. This research is distinct from prior research on CO2 emissions in that it examines the moderating role of financial inclusion, which has been ignored by previous studies. Due to the availability of data, this study examines the period between 2000 and 2023, focusing on the E-7 countries. Environmental pollution is measured using CO2 emissions and is the dependent variable quantified in MtCO2e (Million tonnes). The data for CO2 emissions is obtained from the GCA database (GCA 2023). CO2 is a new metric of carbon emission put forward by Peters et al. (2012) that takes into account international trade. This is based on the idea that CCO2 emissions attribute greenhouse gases to the final consumer of goods and services, regardless of where those emissions were produced, while production-based emissions assign all emissions to the geographical location where they are produced. RE is measured as the percentage of renewable energy to total consumption. Energy intensity, natural resources rents, and economic development (calculated by gross domestic product, GDP) are obtained from the World Bank database. In this study, we constructed a composite index using PCA analysis from five components: Number of ATMs per 100,000 adults (NATM), Branches of commercial banks per 100,000 adults, Number of commercial bank institutions, Outstanding deposits with commercial banks (% of GDP), and Outstanding loans from commercial banks (% of GDP). We used these five different components to measure the FI of E-7 countries, and this will provide us with a score that effectively measures the extent of FI in the E-7 nations. To address data quality concerns, all the variables (except for the financial inclusion index) were log-transformed to mitigate issues of non-linearity, heteroscedasticity, and to facilitate the interpretation of coefficients as elasticities. We centered the financial inclusion (FI) and renewable energy (RE) variables prior to creating their interaction term to reduce multicollinearity between the main effects and the product term, following standard econometric practice for interaction models (Brambor et al. 2006; Iacobucci et al. 2016). Whereas, only descriptive statistics for FI and RE were calculated using log-transformed values, rather than the centered variables. We chose not to impute missing values, but instead, we utilized complete observations to preserve data integrity. Table 1 gives the data description and sources.

Table 1 Variable descriptions and data sources.
Full size table

The selection of E-7 economies is particularly relevant for this analysis as these countries collectively represent emerging market dynamics where rapid financial sector expansion coincides with critical energy transition decisions. These economies - Brazil, China, India, Indonesia, Mexico, Russia, and Turkey - account for approximately 30% of global GDP in purchasing power parity terms and are projected to overtake the G-7 economies by 2050 (PWC 2017). According to the Global Carbon Budget data, these nations collectively accounted for 50.05% of global fossil CO2 emissions in 2023, representing a majority of world emissions (Friedlingstein et al. 2025). This includes China (31.5%), India (8.1%), Russia (4.7%), Indonesia (1.7%), Brazil (1.3%), Mexico (1.3%), and Turkey (1.14%), demonstrating a 30-fold variation between the highest and lowest emitters. This emission share has increased from 48.21% in 2021, underscoring its growing climate significance. The diverse emission trajectories - with rapid growth in India (+13.0%), Indonesia (+18.4%), and Mexico (+18.5%), moderate growth in China (+3.7%) and Russia (+3.4%), and declining emissions in Brazil (−0.5%) and Turkey (−3.2%), provide an ideal heterogeneous setting for our quantile regression analysis. The E-7 economies capture the development-environment dynamics essential for examining how financial inclusion moderates the renewable energy-emissions nexus.

Econometric model

The core model of this research can be described as:

$$\begin{array}{l}{C}{{\rm{O}}2}_{{\rm{i}},{\rm{t}}}\,=\,{\vartheta }_{0}+{\vartheta }_{1}{{RE}}_{i,t}+{{\vartheta }_{2}{{FI}}_{i,t}+\vartheta }_{3}{{FI}* {RE}}_{i,t}+{\vartheta }_{4}\,{{GDP}}_{i,t}\\\qquad\quad\quad\,\,\,+\;{\vartheta }_{5}{{EI}}_{i,t}+{\vartheta }_{6}{{NR}}_{i,t}+{\varepsilon }_{i,t}\end{array}$$
(1)

In the model, CCO2 is consumption-based CO2 emissions (CO2), RE is renewable energy, FI is financial inclusion, GDP is economic growth, EI shows energy intensity, NR is natural resources rents, ϑ1, ϑ2, ϑ3, ϑ4, ϑ5, and ϑ6 are parameters, and ε is the error term.

Analytical framework

First, we conducted a comprehensive descriptive statistic to assess the distributional properties of the data, focusing on skewness and kurtosis to evaluate normality. Before assessing the stationarity of the data, this study analyzes the data for cross-sectional dependency (CD) and slope heterogeneity (SHT). The two analyses help select the appropriate unit root test and subsequent econometric techniques. For this purpose, we employed the method proposed by Pesaran and Yamagata (2008) to examine SHT and Pesaran’s (2004) method to evaluate CD analysis. We used the Im et al. (2003) unit root analysis, which accounts for heterogeneous panels. As fixed and random effects regression models cannot account for CD in error terms and non-normal distributions revealed in our preliminary analysis, we adopt cointegration analysis approaches using both Kao (1999) and Pedroni (2004) tests to examine the long-run relationships among the variables.

Given the significant non-normality identified in our descriptive statistics and the presence of both CD and SHT, we employed the Method of Moments Quantile Regression (MMQR) proposed by Machado and Silva (2019). The MMQR approach is particularly suitable for this study analysis as it provides robust estimates across different quantiles of the conditional distribution, addressing the non-normality issues while capturing heterogeneous effects. The MMQR model can be expressed as:

$${Q}_{{Y}_{{it}}}\left(\tau ,|,{X}_{{it}}\right)=\left({\alpha }_{i}+{\delta }_{i}q\left(\tau \right)\right)+{X}_{{it}}^{{\prime} }\left(\beta +\gamma q\left(\tau \right)\right)$$
(2)

where \({Q}_{{Y}_{{it}}}\left(\tau ,|,{X}_{{it}}\right)\) denotes the conditional quantile of CO2 emissions for country \(i\) at time \(t,{\alpha }_{i}\) represents the individual fixed effects, \({\delta }_{i}\) captures the individual scale effects, \({X}_{{it}}\) is the vector of explanatory variables (GDP, NR, EI, FI, RE), \(\beta\) represents the location parameters (average effects), \(\gamma\) represents the scale parameters (distributional effects), and \(q\left(\tau \right)\) is a transformation of the quantile \(\tau\).

The MMQR estimator provides both location and scale effects, where the location function captures the average relationship between the dependent and independent variables:

$${Location}={\alpha }_{i}+{X}_{{it}}^{{\prime} }\beta$$
(3)

And the scale function captures how this relationship varies across the distribution:

$${Scale}={\delta }_{i}+{X}_{{it}}^{{\prime} }\gamma$$
(4)

For Model 2, which incorporates the interaction effect between financial inclusion and renewable energy, the specification becomes:

$$\begin{array}{ll}{Q}_{{CO}{2}_{{it}}}\left(\tau \right)=\left({\alpha }_{i}+{\delta }_{i}q\left(\tau \right)\right)+{\beta }_{1}{GD}{P}_{{it}}+{\beta }_{2}N{R}_{{it}}+{\beta }_{3}E{I}_{{it}}+{\beta }_{4}F{I}_{{it}}\\\qquad\qquad\quad\;\,\,+\,{\beta }_{5}R{E}_{{it}}+{\beta }_{6}\left({RE}\times {FI}\right){it}+{\gamma }^{{\prime} }{Xitq}\left(\tau \right)\end{array}$$
(5)

The quantile-specific effects at any given quantile \(\tau\) can be computed as:

$$\partial {Q}_{{CO}{2}_{{it}}}\left(\tau \right)/\partial {X}_{{kit}}={\beta }_{k}+{\gamma }_{k}q\left(\tau \right)$$
(6)

where \(k\) indexes the explanatory variables. This formulation allows us to examine how the marginal effects of each variable change across the conditional distribution of CO2 emissions.

To ensure the robustness of our MMQR findings, we employed the Recentered Influence Function (RIF) regression with high-dimensional fixed effects following Firpo et al. (2009). The RIF regression directly models the influence of covariates on specific quantiles of the unconditional distribution, providing an alternative perspective on distributional effects. The RIF for quantile \(\tau\) is defined as:

$${RIF}\left({Y;}{q}_{\tau },{F}_{Y}\right)={q}_{\tau }+\frac{\tau -1\left\{Y\le {q}_{\tau }\right\}}{{f}_{Y}\left({q}_{\tau }\right)}$$
(7)

where \({q}_{\tau }\) is the \(\tau\)-th quantile of the distribution \({F}_{Y}\), and \({f}_{Y}\left({q}_{\tau }\right)\) is the density of \(Y\) at \({q}_{\tau }\). The RIF regression then estimates:

$$E[{RIF}\left({Y}_{{it}};{q}_{\tau },{F}_{Y}\right)|{X}_{{it}}]={X}^{{\prime} }{it}{\beta }_{\tau }$$
(8)

This approach allows us to interpret the coefficients as the marginal effects of the explanatory variables on the unconditional quantiles of CO2 emissions, complementing the conditional quantile effects from MMQR. The combination of MMQR and RIF regression provides a comprehensive framework for analyzing the heterogeneous relationships between financial inclusion, renewable energy, and CO2 emissions across the emission distribution in E-7 economies, while effectively addressing the methodological challenges posed by non-normality, cross-sectional dependence, and slope heterogeneity in our panel data.

Results

In this section, we present the outcomes of the previously described econometric techniques. Table 2 presents the descriptive statistics for all variables examined in this study. The results reveal substantial departures from normality across all variables except financial inclusion (FI), as evidenced by the adjusted chi-square statistics. CO2 emissions exhibit significant positive skewness (0.983) with a chi-square value of 16.333 (p = 0.000), indicating that the distribution is right-tailed. GDP shows the highest kurtosis (4.418) with a chi-square value of 29.395 (p = 0.000), suggesting a leptokurtic distribution with heavy tails and a sharp peak. Energy intensity (EI) demonstrates moderate positive skewness (0.463), while renewable energy (RE) shows negative skewness (−0.555), indicating a concentration of higher renewable energy consumption values. The financial inclusion index (FI) shows the most normal distribution with a chi-square value of 1.742 (p = 0.419). These significant departures from normality across key variables justify our methodological choice of MMQR over traditional mean-based regression techniques, as MMQR provides more robust and reliable estimates when dealing with non-normal distributions.

Table 2 Descriptive statistics.
Full size table

This study uses the CD-test analysis suggested by Pesaran (2004), and the results are shown in Table 3. The results show that there is strong evidence of CD among all variables. The financial inclusion index (FI) exhibits a high cross-sectional dependence with a CD-statistic of 14.372, followed closely by GDP (15.115), natural resources (NR) with 11.713, and renewable energy (RE) with 4.371. These exceptionally high values indicate that shocks in one E-7 economy have substantial spillover effects on others, suggesting strong economic and financial integration within the economies.

Table 3 Cross-sectional dependence and slope homogeneity analysis.
Full size table

The results of the slope heterogeneity (SHT) tests for Models 1-2 are given in Table 3. For Model 1, which examines the direct effects, the Delta statistic is 4.602, and the adjusted Delta statistic is 6.074. Model 2, incorporating the interaction term between financial inclusion and renewable energy, shows Delta and adjusted Delta statistics of 4.018 and 5.610, respectively. These findings are statistically significant at the 1% level, indicating strong evidence of slope heterogeneity.

To examine the stationarity of the data, we used the method put forward by Im et al. (2003). Table 4 displays the outcomes of the unit root test. The results reveal mixed orders of integration among the variables. CO2 emissions, GDP, natural resources (NR), energy intensity (EI), and renewable energy (RE) all exhibit non-stationarity at levels but become stationary after first differencing, indicating I(1) processes. Financial inclusion (FI) demonstrates stationarity at level I(0) with a test statistic of −2.438 (p = 0.014).

Table 4 Unit root analysis.
Full size table

The panel cointegration analysis was conducted using both Kao and Pedroni tests to ensure robustness, with results presented in Table 5. For Model 1, the Kao tests provide strong evidence of cointegration relationships. The Modified Dickey–Fuller t-statistic of −1.7628 (p = 0.0390) and the Unadjusted modified Dickey–Fuller t-statistic of -2.3054 (p = 0.0106) both indicate significant long-run relationships at the 5% level. The Pedroni tests, which allow for heterogeneous cointegrating vectors, further strengthen these findings with the Modified Phillips–Perron t-statistic of 2.5314 (p = 0.0057) and the Augmented Dickey–Fuller t-statistic of −2.5502 (p = 0.0054), both significant at the 1% level. Model 2, which includes the interaction term (RExFI), demonstrates even stronger cointegration evidence across all test statistics, with the Kao Modified Dickey–Fuller t-statistic improving to −2.6243 (p = 0.0043) and the Pedroni Phillips–Perron t-statistic showing a significant value of −9.9426 (p = 0.0000). These results confirm robust long-run associations between CO2 emissions, GDP, natural resources, energy intensity, financial inclusion, renewable energy, and their interaction term. After establishing the long-term linkages among the factors, we applied the MMQR approach to quantify the heterogeneous effects across different quantiles of the CO2 emissions distribution. Table 6 presents the MMQR results for Model 1.

Table 5 Panel cointegration test results.
Full size table
Table 6 MMQR Model 1 (CO2 = GDP, NR, EI, FI, RE).
Full size table

In Table 6, the results show an increasing pattern of GDP’s impact on emissions from lower to higher quantiles. At the 25th quantile, the coefficient is 0.593, rising to 0.655 at the median, 0.679 at the 75th quantile, and reaching 0.691 at the 90th quantile. This progressive increase suggests that countries with higher emission levels experience a more pronounced impact of economic growth on CO2 emissions. These findings are similar to earlier studies by Ali, Rahaman et al. (2025), and Ayad et al. (2023). Whereas, Iqbal et al. (2023) showed that there is a bidirectional causal relationship between economic growth and carbon emissions. Energy intensity (EI) demonstrates a consistently positive and significant relationship with CO2 emissions across all quantiles. The quantile-specific effects show stability, ranging from 0.526 at the 25th quantile to 0.551 at the 90th quantile. This stable relationship across the distribution suggests that improving energy efficiency remains a critical policy lever for all E-7 countries, regardless of their current emission levels. Natural resources (NR) show no significant effect on CO2 emissions across any quantile. The quantile-specific coefficients range from -0.0603 at the 25th quantile to 0.0110 at the 90th quantile. This finding suggests that natural resource rents do not directly influence emission patterns in E-7 economies because these countries have diversified economies where natural resource rents play a varying but not dominant role in determining emission trajectories.

Financial inclusion (FI) exhibits a positive and significant effect on CO2 emissions with a location parameter of 0.0851 (p < 0.01). Interestingly, the quantile analysis reveals a decreasing pattern of impact as we move toward higher emission quantiles. The coefficient at the 25th quantile is 0.0941, declining to 0.0829 at the median, 0.0785 at the 75th quantile, and 0.0764 at the 90th quantile. This pattern suggests that financial inclusion’s contribution to emissions is more pronounced in countries with lower baseline emissions. This finding indicates that as financial services expand in E-7 economies, they initially contribute more to emissions through increased economic activity and consumption, but this effect moderates as countries reach higher emission levels, due to the emergence of environmentally conscious financial products and services.

The results for renewable energy consumption (RE) demonstrate a significant negative effect on CO2 emissions. The quantile analysis reveals substantial heterogeneity in this relationship. At the 25th quantile, renewable energy shows its strongest emission-reducing effect with a coefficient of −0.283, indicating that a 1% increase in renewable energy consumption reduces emissions by 0.283% for lower-emitting countries. This effect diminishes progressively at higher quantiles, with coefficients of −0.130 at the median, −0.0693 at the 75th quantile, and becoming statistically insignificant at the 90th quantile (−0.0412). This suggests that renewable energy investments are most effective in countries that are still developing their energy infrastructure and have lower baseline emissions. For high-emitting countries, the effectiveness of renewable energy in reducing emissions appears limited due to the lock-in effects of existing fossil fuel infrastructure or the scale of emissions that overwhelm the contribution of renewable sources. These results are in line with the study of Qing et al. (2024). Figure 4 demonstrates the visual representation of Model 1 analysis.

Fig. 4: MMQR Model 1 (CO2 = GDP, NR, EI, FI, RE).
figure 4

Each panel shows coefficient estimates across quantiles for each variable. Source: Author’s own calculation.

Full size image

Table 7 presents the MMQR results for Model 2, which incorporates the interaction term between financial inclusion and renewable energy (RExFI). The inclusion of this interaction term provides crucial insights into how financial development conditions affect the relationship between renewable energy and emissions. The results in Model 2 for GDP, natural resources, energy intensity, financial inclusion, and renewable energy are similar to Model 1, with some variations in magnitude. These consistent findings across both models strengthen the robustness of our results and confirm that the inclusion of the interaction term does not fundamentally change the individual relationships between these variables and CO2 emissions. Our findings contrast with Deng et al. (2024), who found that financial development enhances environmental quality and its interaction with renewable energy protects environmental excellence in N-11 countries, whereas our results reveal that financial inclusion increases emissions and weakens renewable energy’s emission-reducing impact in E-7 economies.

Table 7 MMQR Model 2 (CO2 = GDP, NR, EI, FI, RE and RE*FI).
Full size table

The interaction term (RExFI) reveals a positive and significan effectt, indicating that financial inclusion moderates the emission-reducing effect of renewable energy. The quantile analysis shows that this moderating effect is strongest at the 25th quantile (0.297) and decreases to 0.136 at the 90th quantile. The findings show that while renewable energy reduces emissions and financial inclusion facilitates economic activity that increases emissions, their interaction suggests that financial development may be channeling resources toward carbon-intensive activities rather than supporting the renewable energy transition. The stronger moderation at lower quantiles indicates that this effect is most pronounced in countries just beginning their emission trajectories, where financial systems may not yet have developed green finance mechanisms. Figure 5 visually demonstrates the results for Model 2.

Fig. 5: MMQR Model-2 (CO2 = GDP, NR, EI, FII, RE, RE*FI).
figure 5

Each panel shows coefficient estimates across quantiles for each variable. Source: Author’s own calculation.

Full size image

To ensure the robustness of our MMQR findings, we conducted a Recentered Influence Function (RIF) regression analysis using high-dimensional fixed effects, with results presented in Tables 8 and 9. The RIF regression approach provides an alternative method for examining distributional effects by directly modeling the influence of covariates on specific quantiles of the outcome distribution, offering complementary insights to our MMQR analysis.

Table 8 RIF regression analysis Model 1.
Full size table
Table 9 RIF regression analysis Model 2.
Full size table

The RIF results for Model 1 in Table 8 are similar to MMQR findings while revealing some additional findings. The observed sign reversals reflect distributional heterogeneity, as RIF captures unconditional effects while MMQR estimates conditional effects, with control variables naturally exhibiting varying sensitivities across quantiles. Financial inclusion maintains positive and significant effects at intermediate quantiles (0.185 at Q50, 0.204 at Q75) but loses significance at the extremes. Most importantly, renewable energy exhibits negative coefficients at lower quantiles (−0.208 at Q25, −0.261 at Q50), consistent with our MMQR findings, though becoming insignificant at higher quantiles. The R-squared values ranging from 0.491 to 0.731 indicate a strong model fit across all quantiles, with the highest explanatory power at the 90th quantile. Figure 6 visually represents the RIF regression coefficients.

Fig. 6: RIF regression- Model 1.
figure 6

The plot displays coefficient estimates for each variable (GDP, NR, EI, FI, RE) across three quantiles (Q25, Q50, Q75). Source: Author’s own calculation.

Full size image

The RIF regression results for Model 2 (Table 9) provide crucial validation for our interaction effect findings. The interaction term (RExFI) demonstrates strong positive and significant effects across the first three quantiles, with remarkably stable coefficients of 0.346 (p < 0.01) at the 25th quantile, 0.342 (p < 0.01) at the median, and increasing to 0.468 (p < 0.01) at the 75th quantile. These results are similar to the earlier MMQR findings regarding the moderating role of financial inclusion on renewable energy’s emission-reducing potential. The R-squared values (0.636–0.740) indicate robust model performance, with the interaction term contributing substantially to the explanatory power across all quantiles. These RIF results provide strong robustness to our conclusion that current financial development patterns in E-7 economies may be channeling resources toward carbon-intensive activities rather than supporting the renewable energy transition, highlighting the need for green finance frameworks to align financial inclusion with climate mitigation objectives. Figure 7 visually represents the RIF regression coefficients.

Fig. 7: RIF regression- Model 2.
figure 7

The plot displays coefficient estimates for each variable (GDP, NR, EI, FI, RE, RExFI) across three quantiles (Q25, Q50, Q75). Source: Author’s own calculation.

Full size image

Conclusion and policy recommendations

Despite promising growth and economic success, emerging economies’ increased CO2 emissions from the growth process have made it difficult for them to meet the Sustainable Development Goals (SDGs). Key factors that could reduce CO2 emissions in these countries have been identified in several studies. This study explores the role of FI as a moderator in the correlation between renewable energy and CO2 emissions while controlling for GDP, natural resources, and energy intensity. The analysis of cointegration using Kao and Pedroni tests confirms the long-run linkage between FI, RE, GDP, NR, EI, and CO2 emissions. The results of the MMQR analysis reveal significant heterogeneity across the emission distribution and demonstrate that FI positively moderates the relationship between renewable energy and emissions, thereby weakening the emission-reducing impact of renewable energy. While renewable energy usage shows a significant adverse effect on CO2 emissions, particularly at lower quantiles, this beneficial impact diminishes for higher-emitting countries and is substantially offset by the positive interaction with financial inclusion. The interaction term’s positive coefficient across all quantiles indicates that current financial development patterns may be channeling resources toward carbon-intensive activities rather than supporting the renewable energy transition. Additionally, GDP and energy intensity exhibit positive associations with CO2 emissions across all quantiles, contributing to environmental degradation, with these effects intensifying at higher emission levels.

Based on the results of this study, the E-7 countries face a complex challenge in balancing financial inclusion objectives with climate mitigation goals. Our MMQR analysis reveals that while financial inclusion promotes economic development, it currently contributes to increased CO2 emissions, with this effect being more pronounced at lower emission quantiles. More critically, the positive interaction between FI and renewable energy indicates that financial inclusion is moderating the emission-reducing potential of renewable energy investments, suggesting that expanded financial access is channeling resources toward carbon-intensive activities rather than supporting the green transition.

Given these findings, policymakers in E-7 countries should adopt a nuanced, quantile-specific approach to environmental policy. For countries at lower emission quantiles, where renewable energy shows the strongest emission-reducing effects (−0.283 at Q25), priority should be given to rapidly scaling renewable energy infrastructure while simultaneously developing green finance frameworks within their financial inclusion initiatives. These countries have the opportunity to leapfrog carbon-intensive development pathways if financial resources are properly directed. For higher-emitting countries within the E-7, where renewable energy’s effectiveness diminishes, more comprehensive structural reforms are necessary, including aggressive energy efficiency improvements and industrial restructuring, as energy intensity remains a significant emission driver across all quantiles.

To realign financial inclusion with climate objectives, E-7 countries must urgently implement green finance regulations and incentives. This includes (1) mandatory environmental criteria for financial institutions participating in inclusion programs, (2) preferential lending rates and terms for renewable energy projects accessible to newly banked populations, (3) green bonds and climate-focused financial products specifically designed for retail investors, and (4) capacity building programs that educate financial service providers and users about sustainable investment opportunities. Without such measures, the expansion of financial services will continue to undermine renewable energy efforts, as evidenced by the consistent positive interaction effects across quantiles.

The heterogeneous effects revealed by our analysis also call for differentiated renewable energy strategies. Countries at lower emission levels should focus on large-scale deployment of solar, wind, and bioenergy infrastructure, leveraging their higher emission-reduction potential. Meanwhile, higher-emitting E-7 countries need to combine renewable expansion with aggressive energy efficiency measures, grid modernization, and possibly carbon capture technologies. Given that energy intensity shows stable positive effects on emissions across all quantiles, implementing stringent energy efficiency standards, retrofitting programs, and industrial energy management systems becomes crucial for all E-7 nations. These initiatives should be supported by targeted R&D investments that prioritize both renewable energy innovation and energy-efficient technologies, with funding mechanisms that ensure financial inclusion programs channel resources toward these green innovations rather than traditional carbon-intensive sectors.